Iterative three-dimensional image reconstruction from tomographic projections

Abstract

Four algorithms are adapted to perform direct three-dimensional reconstruction from projections. The algorithms considered are summation, the Algebraic Reconstruction Technique (ART), the Simtfltaneous Iterative Reconstruction Technique (SIRT), and the Iterative Least Squares Technique (ILST). The concept of tomographlc projections is introduced and shown to greatly simplify the calculations. This work represents the first time that these iterative algorithms have been applied to projections other than coaxial. The methods developed can be of benefit in electron microscopy, holographic interferometry, and nuclear medicine. To evaluate these methods an experimental investigation is can'ied out using computer-generated synthetic images. Using SIRT, direct 3-D reconstruction is shown to be superior to serial 2-D reconstruction from coaxial projections when the range of viewing angles is limited. The number of projections required for adequate reconstruction is also considered. Finally, the perf rmance of the algorithms is compared with respect to overall similarity of the reconstruction to the original test object, effects of noise, and computer time and memory requirements.